first kotlin example

src/main/kotlin/edu/mcgill/cstk/experiments/repair/KotlinStatementRepair.kt

@@ -10,6 +10,7 @@ import java.io.File
 import kotlin.time.*
 
 /*
+./gradlew kotlinStatementRepair 2>&1 | grep -v "Parser error:" | grep -v "LATEX"
 ./gradlew kotlinStatementRepair 2>&1 | grep -v "Parser error:"
  */
 
@@ -41,10 +42,12 @@ fun main() {
 
       println("\nTop 100 repairs:\n")
       it.take(100).forEach {
-        println("Δ=${levenshtein(prompt, it) - 1} repair: ${prettyDiffNoFrills(prompt, it)}")
+        println("Δ=${levenshtein(prompt, it)} repair: ${prettyDiffNoFrills(prompt, it)}")
+        println("(LATEX) Δ=${levenshtein(prompt, it)} repair: ${latexDiffSingleLOC(prompt, it)}")
       }
 
-      println("Found ${it.size} valid repairs in ${elapsed}ms, or roughly ${it.size / (elapsed/1000.0)} repairs per second.")
+      println("Found ${it.size} valid repairs in ${elapsed}ms, or roughly " +
+        "${(it.size / (elapsed/1000.0)).toString().take(5)} repairs per second.")
       println("Original string was ${if (contained) "#${it.indexOf(original)}" else "NOT"} in repair proposals!\n")
     }
   }
@@ -109,6 +112,7 @@ fun parallelRepairKotlinStatement(
       val levDiff = levenshtein(prompt, it) - 1
       if (levDiff < bestRepair) {
         println("Δ=$levDiff repair: ${prettyDiffNoFrills(prompt, it)}")
+        println("(LATEX) Δ=$levDiff repair: ${latexDiffSingleLOC(prompt, it)}")
         bestRepair = levDiff
       }
     },

src/main/kotlin/edu/mcgill/cstk/experiments/repair/PythonStatementRepair.kt

@@ -103,7 +103,7 @@ fun repairPythonStatement(
   coarsen = String::coarsenAsPython,
   uncoarsen = String::uncoarsenAsPython,
   synthesizer = satRepair(clock), // Enumerative search
-  diagnostic = { println("Δ=${ levenshtein( prompt, it ) - 1 } repair: ${prettyDiffNoFrills(prompt, it)}") },
+  diagnostic = { println("Δ=${levenshtein(prompt, it) - 1} repair: ${prettyDiffNoFrills(prompt, it)}") },
   filter = { isValidPython() },
 )
 

src/main/kotlin/edu/mcgill/cstk/utils/StringUtils.kt

@@ -1,19 +1,19 @@
 package edu.mcgill.cstk.utils
 
+import ai.hypergraph.kaliningraph.parsing.tokenizeByWhitespace
 import ai.hypergraph.kaliningraph.types.cc
 import com.github.difflib.text.*
 import com.github.difflib.text.DiffRow.Tag.*
 import edu.mcgill.cstk.disk.*
 import edu.mcgill.cstk.experiments.probing.embeddingServer
-import edu.mcgill.cstk.experiments.repair.isValidKotlin
+import edu.mcgill.cstk.experiments.repair.defaultTokenizer
 import info.debatty.java.stringsimilarity.interfaces.MetricStringDistance
 import me.vovak.antlr.parser.*
 import net.sf.extjwnl.data.PointerUtils.*
 import net.sf.extjwnl.dictionary.Dictionary
 import org.antlr.v4.runtime.*
 import org.apache.commons.lang3.StringUtils
 import org.jetbrains.kotlin.lexer.*
-import org.jetbrains.kotlin.spec.grammar.tools.tokenizeKotlinCode
 import spoon.Launcher
 import java.io.File
 import java.net.*
@@ -339,18 +339,18 @@ $summary
 
 \subsection{Original}
 \begin{lstlisting}[language=java]
-${diffString(original, synthetic).first}
+${latexDiffMultilineStrings(original, synthetic).first}
 \end{lstlisting}
 \subsection{Synthetic}
 
 \begin{lstlisting}[language=java]
-${diffString(original, synthetic).second}
+${latexDiffMultilineStrings(original, synthetic).second}
 \end{lstlisting}
 
 \subsection{Variant}
 
 \begin{lstlisting}[language=java]
-${diffString(original, variant).second}
+${latexDiffMultilineStrings(original, variant).second}
 \end{lstlisting}
 
 \subsection{Comment}
@@ -366,7 +366,7 @@ $discrepancy
 %--------
 """.trimIndent().also { println(it) }
 
-fun diffString(old: String, new: String) =
+fun latexDiffMultilineStrings(old: String, new: String) =
   DiffRowGenerator.create()
     .showInlineDiffs(true)
     .ignoreWhiteSpaces(true)
@@ -415,6 +415,22 @@ fun String.visibleLen() =
     .replace(ANSI_GREEN_BACKGROUND,"")
     .replace(ANSI_RESET,"").length
 
+fun latexDiffSingleLOC(original: String, new: String) =
+  DiffRowGenerator.create()
+    .showInlineDiffs(true)
+    .inlineDiffByWord(true)
+    .newTag { l -> if(l) "(*@<begin>" else "<end>@*)" }
+    .build()
+    .generateDiffRows(original.tokenizeByWhitespace(), new.tokenizeByWhitespace())
+    .joinToString(" ") {
+      when (it.tag) {
+        INSERT -> it.newLine.replace("<begin>", "\\hlgreen{").replace("<end>", "}")
+        CHANGE -> it.newLine.replace("<begin>", "\\hlorange{").replace("<end>", "}")
+        DELETE -> "\\hlred{${List(it.oldLine.length){ " " }.joinToString("")}}"
+        else -> it.newLine.replace("<begin>", "").replace("<end>", "")
+      }
+    }.replace("&lt;", "<").replace("&gt;", ">")
+
 // Just print the new line with ASCII colors but no border
 fun prettyDiffNoFrills(original: String, new: String) =
   DiffRowGenerator.create()